eMetrics London 2015 - Getting data visualisation to work

PowerPoint Presentation

Getting Data Visualisation to workSean Burton

@analytdata#eMetrics

1

About meSean Burton [email protected] | @analytdata | analyt.co.uk

I'm passionate about improving customer experience and business value by using a blend of data, technology and psychology.Formerly the Director of Measurement atSeren Design Ltd.A 15 year career covering: eLearning, Content Management Systems, Interaction Design, Product Management, Web Analytics, and Data Visualisation.Extensive experience with FTSE 100 companies across financial, telecommunication, gaming, and retail sectors.

@analytdata#eMetricsWho Ive worked with

@analytdata#eMetrics3

ServicesBlended Measurement SolutionWhat Analyt doesDigital customer intelligence consultancy, specialising in services that optimise the Customer Experience (CX) journey & drive digital business performanceTechnical feasibility AuditsBusiness RequirementsPlanning, Design and Metric MappingImplementation supportReporting, automation and visualisationTraining, optimisation and maintenanceCustomer advocacy and modellingOptional On-going support

Data Analysis & Visualisation

Web & Digital Analytics

UX Design & Accessibility

Data Integration & AutomationPlanningAuditing & ImplementationSupport & Training

Surveys & Voice of Customer

@analytdata#eMetricsData Visualisations

@analytdata#eMetricsJust for Jims wife

@analytdata#eMetricsThe honest truthOver the years Ive produced plenty of bad reports

Ive worked with lots of clients that have tons of terrible reports

Why terrible?


Loads of effort &a complete waste of time


And so, this is a story of what Ive learned

@analytdata#eMetricsThe StartAre you sitting comfortably? Then Ill begin

@analytdata#eMetricsFrustrations of an AnalystMeet Jeff

He spent all his time writing reports and was frustrated that nothing ever seemed to happen

Execs were frustrated because they didnt understand the implications of his reports and so couldnt act on them

He was depressed, had lots of sick days, and was on notice of possible redundancy

@analytdata#eMetricsData visualisation=communication =action

@analytdata#eMetricsA bit of TheoryHow we perceive


PhysicalPrinted?On Screen?On Mobile?Distance / TimeBlack/White or Colour?PhysiologicalMale red/green colour-blindnessRetinal DensityFocal PointsEye StrainPsychologicalQuicker to process vertical & horizontal linesQuicker to process an image when proportions match Golden RatioColour association / BiasInformation overload

@analytdata#eMetricsA bit of theory: Information Overload


Everything should be made as simple as possible but not simpler.


But far too often


The middleThe hard fought journey of our hero





S#!t in

=

S#!t out

@analytdata#eMetricsDashboards: Good KPIs are bermetrics


23

Dashboards: development process


24

Dashboards: Stephen Few & Edward Tufte

Use Sparklines to show trend of metric during selected periodWarning flag icons to indicate a significant negative eventAll metrics provided with period comparators to give contextSynchronised Bullet charts used to show metrics against target

@analytdata#eMetricsDashboards: Example Charity Dashboard

@analytdata#eMetricsUX of dashboard the drunk user test

http://drunkusertesting.com



http://drunkusertesting.com



@analytdata#eMetricsStory TellingKnow your audience include something that they can relate toStart with the pain points put the audience in the shoes of their customersMove onto a narrative that combines hard quantitative metrics with actual qualitative customer feedback Restrict yourself to highlighting only 3 key problems to avoid the audience getting lost keep focusedDashboards should be interactive allowing the user to investigate & explore Add filters or actions to allow for data driven navigationA commentary can be added to highlight known issues or areas of interest. Remember to focus on implications/reasons rather than simply saying metric A went up by X%Tableau provides a story mode allowing an analyst to pre-configure a sequence of dashboard manipulations providing a guided data tourFinish with your solutions to the problems and the expected benefits

@analytdata#eMetricsThe EndOur hero saves the day roll end credits

@analytdata#eMetricsThe End5 Key elements for Dashboard success:RelevanceMake sure youre showing the right stuff to the right person at the right time

ContextTry to ground each metric, by showing the: metric; trend; and comparator (3D Metric Presentation)Also think about other associated metrics

ColourUse sparingly, e.g. only red for alertsDont solely depend on colour to convey meaning couple with an icon, e.g. green up-arrow vs red down-arrow.Consider colour associations, e.g. Blue for Conservative & Red for Labour.

StoryConfigure the dashboard to tell a story. Most people read top-left to bottom-right try to layout metrics with a logical flow and consider adding a commentary.

AestheticBe driven by function and not form. However, tailor your design to your audience, you dont want an exec to be put off simply because the dashboard is ugly!

@analytdata#eMetricsThe End.Finish on a high - point out the opportunities rather than the problems

Ensure you have a feedback loop & iterate reports to ensure value

Remember Jeff?Spend most of his time talking with end-users & builds reports based on what gives most value;Automates basic reporting tasks so he can work on something more interesting;Hes happy, got promoted, and is now a vital part of the decision process;Everyone wins!


[email protected]@analytdata0191 704 2045Wanalyt.co.ukDigital customer intelligence consultancy, specialising in services that optimise customer experience & drive digital business performance


eMetrics London 2015 - Getting data visualisation to work

Data & Analytics